Facial Landmarks Based Drowsy Driver Detection Model with OpenCV and DLib

In the present world, a driver falling asleep while driving is one of the major causes of accidents. It is highly dangerous for the driver as well as the others passing by the vehicle who are prone to accidents. It is possible to detect the early signs of drowsiness, which helps the driver become conscious and not lose control of the vehicle. This paper presents a facial landmark-based drowsy driver detection model using computer vision. It uses the OpenCV module to detect the drowsiness level of a driver. Further DLib modules are used to extract the facial landmarks, process the required eye and mouth parameters, and alert the driver with an alarm. This research consists of three parts. Initially, it captures the live video and feeds it to the computer, where a pre-trained model is used to get the facial landmarks. Secondly, the pre-trained facial landmarks are processed using a machine learning module to calculate the required EAR and MAR ratios. Finally, the alarm system is enabled after finding the state of the driver using the calculated metrics.